Accelerated Gradient-Free Optimization Methods with a Non-Euclidean Proximal Operator
- 6 Downloads
We propose an accelerated gradient-free method with a non-Euclidean proximal operator associated with the p-norm (1 ⩽ p ⩽ 2). We obtain estimates for the rate of convergence of the method under low noise arising in the calculation of the function value. We present the results of computational experiments.
Keywordsaccelerated optimization methods convex optimization non-gradient methods inaccurate oracle non-Euclidean proximal operator prox-structure
Unable to display preview. Download preview PDF.
The work shown in Section 3 was supported by the Russian Science Foundation, project no. 17-11-01027. In the remaining sections, the work of A.V. Gasnikov was funded within the framework of the State Support of the Leading Universities of the Russian Federation “5-100” and was supported by the Russian Foundation for Basic Research, project no. 18-31-20005 mol-a-ved, the work of E.A. Gorbunov was supported by the grant of the President of the Russian Federation MD-1320.2018.1, the work of P.E. Dvurechenskii and E.A. Vorontsova was supported by the Russian Foundation for Basic Research, project no. 18-29-03071 mk.
- 4.Rumelhart, D.E., Hinton, G.E., and Williams, R.J., Learning Representations by Back-Propagating Errors, Nature, 1986, no. 323, pp. 533–536.Google Scholar
- 7.Nikolenko, S., Kadurin, A., and Arkhangel’skaya, E., Glubokoe obuchenie. Pogruzhenie v mir neironnykh setei (Deep Learning: An Immersion in the World of Neural Networks), St. Petersburg: Piter, 2018.Google Scholar
- 11.Gasnikov, A.V., Dvurechenskii, P.E., and Nesterov, Yu.E., Stochastic Gradient Methods with an Imprecise Oracle, Tr. MFTI, 2016, vol. 8, no. 1, pp. 41–91, arXiv preprint arXiv:1411.4218.Google Scholar
- 12.Chopra, P., Reinforcement Learning without Gradients: Evolving Agents using Genetic Algorithms, https://towardsdatascience.com/reinforcement-learning-without-gradients-evolving-agents-usinggenetic-algorithms-8685817d84f
- 14.Allen-Zhu, Z. and Orecchia, L., Linear Soupling: An Ultimate Unification of Gradient and Mirror Descent, arXiv preprint arXiv:1407.1537.Google Scholar
- 15.Dvurechensky, P., Gasnikov, A., and Tiurin, A., Randomized Similar Triangles Method: A Unifying Framework for Accelerated Randomized Optimization Methods (Coordinate Descent, Directional Search, Derivative-Free Method), arXiv preprint arXiv:1707.08486.Google Scholar
- 16.Nemirovskii, A.S. and Yudin, D.B., Slozhnost’ zadach i effektivnost’ metodov optimizatsii (Complexity of Problems and Efficiency of Optimization Methods), Moscow: Nauka, 1979.Google Scholar
- 18.Bogolubsky, L., Dvurechensky, P., Gasnikov, A., Gusev, G., Nesterov, Y., Raigorodskii, A., Tikhonov, A., and Zhukovskii, M., Learning Supervised PageRank with Gradient-Based and Gradient-Free Optimization Methods, in Lee, D.D., Sugiyama, M., Luxburg, U.V., Guyon, I., Garnett, R., Eds., Proc. 30th Int. Conf. on Neural Information Processing Systems (NIPS’16), pp. 4914–4922, Curran Associates, Inc., 2016, arXiv:1603.00717.Google Scholar
- 19.ACDF method Python code. https://github.com/evorontsova/ACDF
- 20.Gasnikov, A.V., Effektivnye chislennye metody poiska ravnovesii v bol’shikh transportnykh setyakh (Efficient Numerical Methods for Finding Equilibria in Large-Scale Transportation Networks), Doctoral Dissertation, Moscow: MFTI, 2016, arXiv preprint arXiv:1607.03142.Google Scholar